GeoSTAR A New Approach for a Geostationary Microwave Sounder Bjorn Lambrigtsen 13th International TOVS Study Jet Propulsion Laboratory California Institute of Technology Conference Ste. Adèle, Canada October 28 to November 4 2003
Credits Bjorn Lambrigtsen Bjorn.Lambrigtsen@jpl.nasa.gov Jet Propulsion Laboratory California Institute of Technology This work was carried out at the Jet Propulsion Laboratory, California Institute of Technology under a contract with the National Aeronautics and Space Administration
Summary GeoSTAR is a microwave sounder intended for GEO deployment Also suitable for MEO Functionally equivalent to AMSU Tropospheric T-sounding @ 50 GHz with 50 km resolution Primary usage: Cloud clearing of IR sounder Secondary usage: Stand-alone soundings Tropospheric q-sounding @ 183 GHz with 25 km resolution Primary usage: Rain mapping Secondary usage: Stand-alone soundings Using Aperture Synthesis Also called Synthetic Thinned Array Radiometer (STAR) Also called Synthetic Aperture Microwave Sounder (SAMS)
Why? GEO sounders complement LEO sounders LEO: Global coverage, but poor temporal resolution; high spatial res. is easy GEO: High temporal resolution and coverage, but only hemispheric non-polar coverage; high spatial res. is hard Requires equivalent measurement capabilities as now in LEO: IR + MW Enable full sounding capability from GEO Complement primary IR sounder with matching MW sounder Until now not feasible due to very large aperture required (~ 4-5 m dia.) Microwave provides cloud clearing information Requires T-sounding through clouds Must reach surface under all atmospheric conditions Stand-alone IR sounders are only marginally useful Can sound down to cloud tops ( clear channels ) Can sound in clear areas ( hole hunting ) Clear scenes make up < 2% globally at AMSU resolution (50 km) As clear criteria are relaxed, retrieval errors grow Both exclude active-weather regions & conditions In particular: The all-important boundary layer is poorly covered
Functionality & Benefits of GeoSTAR Soundings Full hemisphere @ 50/25 km every 30-60 min (continuous) - initially, but easily improved Cloudy & clear conditions Complements any GOES IR sounder Enables full soundings to surface under cloudy conditions Rain Full hemisphere @ 25 km every 30 min (continuous) - initially, but easily improved Measurements: scattering from ice caused by precipitating cells Real time: full hemispheric snapshot every 30 minutes or less Synthetic aperture approach Feasible way to get adequate spatial resolution from GEO Easily expandable: aperture size, channels -> Adaptable to changing needs Easily accommodated: sparse array -> Can share real estate with other subsystems Above all: No moving parts -> Minimal impact on host platform & other systems
Background GeoSTAR based on GEO/SAMS (1999): One of 4 innovative concepts selected for NMP/EO-3 Study Medium-scale space demo @ 50 GHz, T-sounding only Phase A completed (cost $0.75M) - 9/99 Projected mission cost: $87M (with reserves) Projected payload development cost: $36M (with reserves) Not selected for implementation (GIFTS selected instead) Proto-GeoSTAR: Ground demo now being developed Sponsored by NASA s Instrument Incubator Program (IIP) Similar to GEO/SAMS: small-scale proof-of-concept ground demo @ 50 GHz Projected cost: ~$3M JPL teaming with GSFC (Piepmeier) & U. Mich. (Ruf)
GeoSTAR System Concept Concept Sparse array employed to synthesize large aperture Cross-correlations -> Fourier transform of Tb field Inverse Fourier transform on ground -> Tb field Array Optimal Y-configuration: 3 sticks; N elements Each element is one I/Q receiver, 3λ wide (2 cm @ 50 GHz) Example: N = 100 Pixel = 0.09 50 km at nadir (nominal) One Y per band, interleaved Other subsystems A/D converter; Radiometric power measurements Cross-correlator - massively parallel multipliers On-board phase calibration Controller: accumulator -> low D/L bandwidth Receiver array Resulting uv samples Example: AMSU-A ch. 1
Aperture Synthesis Is Not New Very Large Array (VLA) at National Radio Astronomy Observatory (NRAO) In operation for many years
Others Are Developing STAR for Space ESA s Soil Moisture and Ocean Salinity (SMOS) L-band system under development - Launch in 2006-2008
What GeoSTAR Measures Visibility measurements Essentially the same as the spatial Fourier transform of the radiometric field Measured at fixed uv-plane sampling points - One point for each pair of receivers Both components (Re, Im) of complex visibilities measured Visibility = Cross-correlation = Digital 1-bit multiplications @ 100 MHz Visibilities are accumulated over calibration cycles > Low data rate Calibration measurements Multiple sources and combinations Measured every 20-30 seconds = calibration cycle Interferometric imaging All visibilities are measured simultaneously - On-board massively parallel process Accumulated on ground over several minutes, to achieve desired NEDT 2-D Fourier transform of 2-D radiometric image is formed - without scanning Spectral coverage Spectral channels are measured one at a time - LO tunes system to each channel
Calibration GeoSTAR is an interferometric system Therefore, phase calibration is most important System is designed to maintain phase stability for tens of seconds to minutes Phase properties are monitored beyond stability period (e.g., every 20 seconds) Multiple calibration methods Common noise signal distributed to multiple receivers > complete correlation Random noise source in each receiver > complete de-correlation Environmental noise sources monitored (e.g., sun s transit, Earth s limb) Occasional ground-beacon noise signal transmitted from fixed location Other methods, as used in radio astronomy Absolute radiometric calibration One conventional Dicke switched receiver measures zero baseline visibility Same as Earth disk mean brightness temperature (Fourier offset) Also: compare with equivalent AMSU observations during over/under-pass The Earth mean brightness is highly stable, changing extremely slowly
GeoSTAR Data Processing On-board measurements Instantaneous visibilities: high-speed cross-correlations Accumulated visibilities: accumulated over calibration cycles Calibration measurements On-ground image reconstruction Apply phase calibration: Align calibration-cycle visibility subtotals Accumulate aligned visibilities over longer period > Calibrated visibility image On-ground image reconstruction Inverse Fourier transform of visibility image, for each channel Complexities due to non-perfect transfer functions are taken into account On-ground geophysical retrievals Conventional approach Applied at each radiometric-image grid point
Technology Development MMIC receivers Required: Small (2 cm wide slices @ 50 GHz), low power, low cost Status: Receivers off-the-shelf @ < 100 GHz; Chips available up to 200 GHz Correlator chips Required: Fast, low power, high density Status: Real chips developed for IIP & GPM; Now 0.5 mw per 1-bit @ 100 MHz Calibration Required: On-board, on-ground, post-process Status: Will implement & demo GEO/SAMS design in Proto-GeoSTAR System Required: Accurate image reconstruction (Brightness temps from correlations) Status: Will demonstrate capability with Proto-GeoSTAR Related efforts: Rapidly maturing approach & technology European L-band SMOS now in Phase B; to be launched ~2006-8 NASA X/K-band aircraft demo (LRR): candidate for GPM constellation NASA technology development efforts (IIP, etc.); various stages of completion
GeoSTAR vs. Real-Aperture Approach Feature GeoSTAR Real-Aperture Aperture size Any size Limited Scanning No scanning Mechanical scanning Spatial coverage Full disk Limited Spectral coverage One array: one band One antenna: all bands Accommodation Easy Difficult Power consumption Now: high; Soon: med. Moderate Platform disturbance None High
Science & Algorithms Rain: New methodology @ sounder frequencies Requires 1 band @ 183 GHz; additional sounding bands are advantageous Advantage: High freq. High res. @ small aperture Algorithms being developed for EOS Aqua/AIRS by Staelin (MIT) Not yet mature - expect mature in ~ 1-2 yrs Being considered to complement GPM Measures snowfall as well as rain: unique capability Soundings: Existing methodology Tropospheric T-sounding requires 1 band @ 50 GHz (4-5 AMSU channels) Full T/q-sounding requires 2 bands @ 50 + 183 GHz (+ windows) Use algorithms developed for AMSU Mature - little further development needed
GeoSTAR Prototype Development Objectives Technology risk reduction Develop system to maturity and test performance Evaluate calibration approach Assess measurement accuracy Small, ground-based 24 receiving elements - 8 (9) per Y-arm Operating at 50-55 GHz 4 tropospheric AMSU-A channels: 50.3-52.8-53.71/53.84-54.4 GHz Implemented with miniature MMIC receivers Element spacing as for GEO application (3 λ) FPGA-based correlator All calibration subsystems implemented
GeoSTAR Prototype Development C/L FEM I&Q I & Q digitizer / / multiplexer LVDS out LVDS out clock control in control in clock correlator correlator wr-15 hybrid bias LO coax pwr ϕ-shift LO control x-face ctl ctl pwr temp. & engineering data subsys data subsys PC power
QuickTime and a TIFF (Uncompressed) decompressor are needed to see this picture. QuickTime and a TIFF (Uncompressed) decompressor are needed to see this picture. QuickTime and a TIFF (Uncompressed) decompressor are needed to see this picture. GEOSTAR GEOSTATIONARY SYNTHETIC THINNED APERTURE RADIOMETER Proto-GeoSTAR Antenna Array Parabolic Potter Horn Gold Plated Copper Knife Edge (0.5 mm) Waveguide transition to WR-15 Calibration distribution Downconverter (0.74x2.0x3.0) Digitizer 6.61" 0.826" backplane support (14.0 x 14.0) 2.14 Y-Array of 24 Horns Prototype 50-GHz Receiver
Calibration Error Budget Individual errors causing equal contribution to overall image-nedt of 1.0 K Array size = delta-t T sys Bτ 50x50 0.0076 0.32 0.19 1.7 0.17 200x200 0.0019 0.32 0.19 3.5 0.17 Additive noise needs to be smaller for larger arrays (same goes for null offsets). Gain and phase tolerances are relaxed for larger spacings, so large arrays have ~ same requirements as small array. Antenna pattern tolerances are not changed by array size.
Roadmap Prototype: 2003-2006 Functional system expected ready in < 1 year Fully characterized in < 2 years Further technology development: 2005-2008 Develop efficient radiometer assembly & testing approach Migrate correlator design & low-power technology to rad-hard ASICs Expect power consumption to reach 0.1 mw per correlator in this time frame Overall power consumption is then trivial: < 100 W for the entire T/q-sounding correlator Develop signal distribution, thermal control & other subsystems. Space demo: 2008-2012 Ready for Phase B in 2008 Ready for launch in 2012
The GeoSTAR Team Bjorn Lambrigtsen (JPL) William Wilson (JPL) Todd Gaier (JPL) Alan Tanner (JPL) Chris Ruf (U. Mich.) Jeff Piepmeier (GSFC) Principal Investigator Task Manager MMIC radiometers System Engineer Correlators & electronics Correlator subsystem & testing Shyam Bajpai (NOAA) James Shiue (GSFC) Science advisory board Science advisory board